Many protein-protein interactions are mediated by short, linear motifs that bind selectively to modular, conserved protein domains. It is a long-standing goal of modern biology to understand the molecular recognition code for these types of interactions, because this would help elucidate mechanisms of signal transduction and predict cellular localization, targeting, and substrate selectivity of proteins that contain peptide-recognition domains. The ability to re-design domain-peptide interactions would lead to protein- interaction inhibitors useful for research and therapy. Recent advances in protein and peptide library screening and sequencing provide exciting opportunities to measure domain-peptide interactions in high throughput. In parallel with such measurements, computational methods are needed to interpret screening results and build models that can be used for protein interaction prediction and design. This proposal describes an integrated experimental/computational program for measuring, modeling and designing interactions mediated by structurally conserved EVH1 domains that bind to short, proline-rich motifs. The proposed studies focus on Ena/VASP family proteins critical for regulation of the actin cytoskeleton. Three human paralogs, Mena, VASP and EVL, act as scaffolds to assemble complexes at the ends of actin fibers. Ena/VASP proteins are critical for the formation of filopodia and lamellipodia and for regulation of actin- based cell motility; they also have newly discovered roles in neural development. Mena, and particularly its isoform Mena invasive, control cancer cell invasion and promote resistance to chemotherapy; the EVH1 binding activity of Mena is therefore an attractive therapeutic target. The peptide binding determinants for EVH1 domains are poorly understood. Short consensus binding motifs have been identified in earlier work but are not sufficient to predict which sequences will bind, or with what paralog specificity. Sequence that flanks known motifs modulates binding to EVH1 domains, but this phenomenon has not been systematically studied. In three specific aims, we propose to (1) experimentally identify Ena/VASP EVH1 domain-binding peptides in the human proteome, determine the sequence requirements for binding, and postulate new biologically relevant interaction partners, (2) apply computational structural modeling and experimental structure determination to understand the mechanisms of binding specificity, including the contributions of core-motif and flanking sequences, and (3) design high affinity and paralog- and/or isoform-selective inhibitors of EVH1 domain interactions. Deeper insight into how EVH1 domains engage their partners will advance our understanding of how these proteins contribute to cell motility, including in invasive cancer cells. Inhibitor design will provide reagents for perturbing Ena/VASP protein functions, including the role of Mena in regulating local mRNA translation in neurons. This work will establish a path for mapping and inhibiting domain-peptide interactions that can be applied to many other peptide-recognition domains.